Project 3.3 - Smart Load Control and Grid Friendly Appliances

147 Pages·2003·1.15 MB·English

PNNL-14342

Final Report for the Energy Efficient and
Affordable Small Commercial and
Residential Buildings Research Program

Project 3.3 - Smart Load Control and
Grid Friendly Appliances

M. Kintner-Meyer R. Guttromson
D. Oedingen S. Lang

July 2003

Prepared for the U.S. Department of Energy
under Contract DE-AC06-76RL01830
DISCLAIMER

This report was prepared as an account of work sponsored by an
agency of the United States Government. Neither the United
States Government nor any agency thereof, nor Battelle
Memorial Institute, nor any of their employees, makes any
warranty, express or implied, or assumes any legal liability or
responsibility for the accuracy, completeness, or usefulness of
any information, apparatus, product, or process disclosed, or
represents that its use would not infringe privately owned
rights. Reference herein to any specific commercial product,
process, or service by trade name, trademark, manufacturer, or
otherwise does not necessarily constitute or imply its
endorsement, recommendation, or favoring by the United States
Government or any agency thereof, or Battelle Memorial
Institute. The views and opinions of authors expressed herein do
not necessarily state or reflect those of the United States
Government or any agency thereof.

PACIFIC NORTHWEST NATIONAL LABORATORY
operated by
BATTELLE
for the
UNITED STATES DEPARTMENT OF ENERGY
under Contract DE-AC06-76RL01830

Printed in the United States of America

Available to DOE and DOE contractors from the
Office of Scientific and Technical Information,
P.O. Box 62, Oak Ridge, TN 37831-0062;
ph: [phone]
fax: [phone]
email: [email protected]

Available to the public from the National Technical Information Service,
U.S. Department of Commerce, 5285 Port Royal Rd., Springfield, VA 22161
ph: [phone]
fax: [phone]
email: [email protected]
online ordering: http://www.ntis.gov/ordering.htm
This document was printed on recycled paper.

(8/00)
Final Report for the Energy Efficient and
Affordable Small Commercial and Residential
Buildings Research Program

Project 3.3 – Smart Load Control and Grid
Friendly Appliances

M. Kintner-Meyer

R. Guttromson
D. Oedingen
S. Lang

July 2003

Prepared for
the U.S. Department of Energy
under Contract DE-AC06-76RL01830

Pacific Northwest National Laboratory
Richland, Washington 99352
Executive Summary
This report summarizes the results of research initiated in April 2000 under funding from the
California Energy Commission and co-funding by the U.S. Department of Energy. The objective
of this project was to develop, implement, and test new methods for detecting pre-cursors of
impending problems in the California electric power grid. The approach pursued in the project
utilized information that is measurable at the wall outlet anywhere in the California. The
approach deliberately focused on methods that do not require communication from an outside
source, but rather function fully autonomously by relying on a local frequency sensor that
measures the frequency of the alternating current (AC) power supply at the wall outlet and some
control intelligence that can ultimately be implemented at low cost in commonly used appliances
for homes and businesses.

During the course of the project, two load controller prototypes were developed, built, and tested.
The first load controller prototype responded to under-frequency events and rapid decay in the
grid frequency. The controller was based on a personal computer (PC) platform with an
Microsoft DOS operating system. The second load controller prototype was used for the
statistical and spectral analysis of historic frequency data of known grid events. It was based on a
PC with a Linux operating system that provided real-time controller capability as well as
processing historic data read from a data file.

The first controller was designed to react to a grid event and then trigger a load to trip off line.
We demonstrated the prototype in the laboratory. An under-frequency load shedding scheme
implemented at end-use devices and appliances has great potential value associated with its ability
to displace reserve generation capacity. This reserve capacity is required to be available during
fast responses of unplanned generation and transmission outages. Instead of utilizing generation
to correct a frequency error, control of loads could be used to achieve the same effect. Thus, the
economic value of a frequency responsive load controller would be similar to that of spinning
reserves.

In an attempt to extend the responsive nature of the first controller prototype, precursors of
impending grid problems specific to California were explored on which to base the development
of a more advanced autonomous load controller design. After consultation with transmission
planning engineers of the CAISO, we analyzed the following two grid problems relevant for
California: 1) dynamic stability problems during high power imports into Southern California
from East of the Colorado River based on the SCIT (Southern California Import Transmission)
nomogram, and 2) voltage stability problems in the San Francisco Bay Area during heavy AC/DC
North-to-South power flows based on the T-116B nomogram. For these problems, we explored
pre-contingency detection methods that were intended to trigger a load reduction in advance of an
impending problem. We used detailed dynamic simulations of the Western power grid (WECC)
and a simplified single-input-single-output model for selected locations in California to explore
signatures in the AC frequency signal for high-stress and low-stress cases. We defined the low-
stress case as a grid condition in which standard CAISO operating procedures were observed.
The high-stress case was defined as a hypothetical case, in which the system was operated outside
CAISO safe operation conditions.

ii i
For the dynamic stability (SCIT) cases, the simulation results revealed recognizable differences
between the high- and low-stress cases in the frequency signal and its autospectrum for different
locations throughout California. This finding gave rise to the formulation of a set of hypotheses
that identified distinct differences in the dynamic behavior of the grid frequency as the power
system transitions from a low-stress to a high-stress condition. The hypotheses postulated were:

1. Higher standard deviation in the frequency signal for the high stress case
2. Higher min-max range in the frequency signal for the high stress case
3. Higher amplitude in the autospectrum of the frequency signal for the high-stress case.

Contrary to the dynamic instability cases, the simulation of the voltage stability problems
furnished results that revealed no differences in the dynamic behavior of the power system
between the high- and the low-stress cases. These results led to the conclusion that the dynamic
analysis approach appears not appropriate for voltage stability problems.

To test the hypotheses postulated, historic data representing two distinct grid events were
analyzed. The first data set represented the WECC breakup of August 10, 1996, that caused
wide-spread outages in the western region. The other data set (dated October 8, 2002)
represented a transmission line trip followed by some remedial action and scattered load loss.

The results of the data analysis did not support our hypotheses. Finding some historic data that
are representative of low- and high-stress conditions was difficult. The randomness and
magnitude of constantly changing loads and adjustments by generators to meet the demand,
coupled with the randomness of the unplanned outages, which cause changes in the topology of
the network, makes it very difficult, if not impossible, to definitively declare a state of the power
system as low stress. Even during periods at night, when the load tends to be lower than during
the day, it is not obvious that the system attains a low or lower-stress state. Transmission outages,
planned or unplanned, may pose a difficult burden on transmission engineers to keep the system
in stable and safe condition. Because of the inherent inability to establish a state of low stress as a
reference case, it became difficult during this analysis of historic data to detect the transition from
a safe condition to that of an impending problem.

A necessary requirement for an effective detection technology is to recognize system conditions
as the power system approaches dangerously close the edge of stable and safe operating
conditions. Because of the complexity of the power system, the edge of safe operations is a
moving target and depends on load conditions and network topology and thus may change from
hour to hour.

As a result of this data analysis, it appears questionable whether the chosen approach will be
successful in the long-run. The major obstacle for this approach is the necessity to establish a
reference scenario that would represent safe grid operating conditions. To establish this, a large
series of the conditions needs to be analyzed to become familiar with the spectrum of variability
for each indicator to establish signatures or patterns for impending problems.

An alternative approach, if feasible, could potentially lead to a promising detection of dynamic
instability of the power system. This alternative approach focuses on determining the transfer
iv
function that describes the dynamic behavior of the entire power system, from which the standard
stability analysis methods can be applied. So far, no one has successfully established a power
system transfer function of sufficient accuracy with which to perform a meaningful stability
analysis.

With the insights gained from the simulation and data analysis, the following recommendations
for additional research are made:

1. Under-frequency load control could provide an important grid reliability enhancement.
Although reactive in its response, an under-frequency load control strategy with frequency
responsive appliances and devices could provide reserves that are currently furnished by
generators that are either already spinning or that can be ramped up in their output.

2. To enhance fundamental understanding of the stability characteristics of the power system, it
is recommended that system identification techniques be used to approximate a real-time
transfer function of the entire power system. If a real-time system transfer function of
sufficient accuracy can be established, it would enable the use of standard stability analysis
tools for determining distance to the stability edge.

3. For dealing with voltage stability problems, we recommend the use of under-voltage relays of
induction motors, as found in compressor motors for air-conditioning systems and other
appliances. The under-voltage protection prevents motor stalling caused by decreasing
voltage as a result of a line fault or high system loading. The stalling of induction motors
perpetuates the decreasing voltage to a point, where the voltage may drop sharply and quickly
and propagate through the distribution systems as other electric motors reach the same
conditions.

v
v i
Table of Contents
Executive Summary.......................................................................................................................iii
1 Introduction................................................................................................................................1
2 Definition of Grid Stress............................................................................................................3
3 Development of a 1st Generation Load Controller..................................................................5
3.1 Objective of First Generation Load Controller....................................................................5
3.1.1 What Electric Grid Events Can Be Detected in the Grid Frequency?..........................5
3.1.2 Design Considerations for Local Frequency Measurements........................................7
3.2 Design and Implementation of the 1st Generation Load Controller.....................................7
3.2.1 Block Diagram of Simplified Load Controller.............................................................7
3.2.2 Frequency Sensor and Controller Description..............................................................8
3.2.3 Prototype of 1st Generation Load Controller...............................................................9
3.2.4 Software Description...................................................................................................11
3.3 Load Controller Capabilities..............................................................................................12
3.4 Value of 1st Generation Load Controller............................................................................12
4 Development of Data Analysis Platform................................................................................13
4.1 Introduction........................................................................................................................13
4.2 Analysis Tool Platform.......................................................................................................13
4.2.1 Hardware.....................................................................................................................14
4.2.2 Software......................................................................................................................14
4.3 Reading Real-Time Data....................................................................................................14
4.4 Algorithms..........................................................................................................................15
4.4.1 Development of Spectrum of a Signal........................................................................15
4.4.2 Spectral Bands.............................................................................................................15
4.4.3 Integral........................................................................................................................16
4.4.4 Maxima.......................................................................................................................16
4.4.5 Sharpness Detection....................................................................................................16
5 Analysis of Grid Stress............................................................................................................19
5.1 Motivation for Enhanced Load Controller for Detection of Grid Stress............................19
5.2 Dynamic Stability Issues....................................................................................................19
5.2.1 Establishment of Grid Stress Conditions - Definition................................................21
5.2.2 Analysis of Grid Stress in the Power System.............................................................22
5.2.3 Findings.......................................................................................................................29
5.2.4 Conclusions from Analysis of Simulation of Dynamic Stability................................30
5.3 Voltage Stability Conditions ..............................................................................................30
5.3.1 Approach.....................................................................................................................31
5.3.2 Findings and Conclusions...........................................................................................31
6 Analysis of Grid Events...........................................................................................................35
6.1 Results for the October 8, 2002, Disturbance.....................................................................35
6.1.1 Time Domain..............................................................................................................35
6.1.2 Magnitude of Maxima in Spectrum............................................................................36
vi i
6.1.3 Standard Deviation......................................................................................................37
6.1.4 Sharpness of Maxima..................................................................................................38
6.1.5 Integral........................................................................................................................39
6.1.6 Histogram of Maxima.................................................................................................40
6.2 Results for August 10, 1996...............................................................................................42
6.2.1 Magnitude of Maxima in Spectrum............................................................................43
6.2.2 Standard Deviation......................................................................................................44
6.2.3 Sharpness of Maxima..................................................................................................45
6.2.4 Integral........................................................................................................................47
6.2.5 Histogram of Maxima.................................................................................................47
7 Conclusions...............................................................................................................................51
8 Recommendations for Future Work......................................................................................53
8.1 Under-frequency Load Control using Grid-Friendly Appliances......................................53
8.2 System Identification Approach of the Electric Power System.........................................56
8.3 Prevention of Stalled Induction Motors.............................................................................57
9 References.................................................................................................................................59
Appendix A: Description of Frequency Sensor........................................................................61
Appendix B: Source Code for Controllers and Analysis Platform Software.........................63
vi ii
List of Figures
Figure 3-1: System Frequency Response to the Loss of a Generator...........................................6
Figure 3-2: Functional Units of a Frequency Sensor Unit............................................................8
Figure 3-3: Block Diagram of 1st Generation Load Controller....................................................9
Figure 3-4: Front- and back-panel of the 1st Generation Load Controller..................................10
Figure 3-5: Inside View of 1st Generation Load Controller Prototype........................................11
Figure 4-1: Schematic of Tool for Data Analysis.......................................................................13
Figure 4-2: Definition of Sharpness at the Maximum of a Discrete Signal................................17
Figure 4-3: Second Derivative Used for Determining the Sharpness at the Maximum of a
Discrete Signal..........................................................................................................17
Figure 4-4: Definition of Sharpness at the Maximum of a Discrete Signal by Determining
the Angle..................................................................................................................18
Figure 5-1: Major Transmission Lines of the Western Electricity Coordinating Council
(WECC)....................................................................................................................20
Figure 5-2: Southern California Import Transmission Nomogram. Locus X inside the
stability boundary is considered low stress. Locus H outside the boundary is
considered the high stress condition.........................................................................21
Figure 5-3: Linear single input/single output model with (1) noise power p(t) input, (2) low
pass filter with 5 Hz break frequency, (3) transfer function, (4) noise response of
frequency f(t), and (5) Fast Fourier Transform or spectrum of frequency f(t).........23
Figure 5-4: Response at Lugo, California, to Chief Joseph Brake event....................................25
Figure 5-5: Autospectrum of System Frequency at Lugo, California. Generated by FFT
with 60 second samples, Hanning squared window, and second-order low-pass
filter breaking at 5 Hz (see Figure 5-3)....................................................................25
Figure 5-6: Response at Vincent, California, to Chief Joseph Brake event................................26
Figure 5-7: Autospectrum of System Frequency at Vincent, California. Generated by FFT
with 60 second samples, Hanning squared window, and second-order low-pass
filter breaking at 5 Hz...............................................................................................26
Figure 5-8: Response at Devers, California, to Chief Joseph Brake event.................................27
Figure 5-9: Autospectrum of System Frequency at Devers, California. Generated by FFT
with 60 second samples, Hanning squared window, and second- order low-pass
filter breaking at 5 Hz...............................................................................................27
Figure 5-10: WECC Breakup of August 10, 1996. Shown is the real power at Malin.
Several events leading to the separation of the interconnected power system are
indicated...................................................................................................................28
Figure 5-11: Spectrum of System Frequency Before and After the Keeler-Alstrom Line
Break, Recorded at Dittmer Control Station, WA....................................................29
Figure 5-12: Transfer Function G (s) at Table Mountain, California, for High-Stress Case
pf
(North California Hydro=100%) and Low-Stress Case (North California
Hydro=70%).............................................................................................................32
Figure 5-13: Spectrum of System Frequency at Table Mountain, California, for High-Stress
(North California Hydro=100%) and Low-Stress (North California
Hydro=70%). These results were generated by FFT with 60 second samples,
ix